In 2012, Magelli launched Apervita to enable health professionals and enterprises to capture, share and use health knowledge. Customers can access the analytics and integrate them into their workflow. One algorithm Mayo is sharing through Apervita helps doctors quickly sift through data and identify patients at risk for sudden cardiac arrest.
“We are sharing our algorithms to empower others to deliver patients the best healthcare,” Dr. Paul Friedman, Mayo's vice chair of cardiovascular medicine, said in an Apervita news release. Friedman was not available for an interview.
Besides Mayo, several other major systems including the University of California at San Francisco and Johns Hopkins have uploaded their algorithms. The most popular analytics focus on early warnings for patient deterioration and reducing hospital readmissions.
Dr. Dana Edelson, a hospitalist at University of Chicago Medicine and co-founder of analytics startup Quant, is working on commercializing an algorithm that uses 30 variables, including vitals signs and demographic data, to predict cardiac arrest, need for intensive care and mortality within 24 hours. The algorithm is 93% accurate, she said. But up to now, disseminating this type of algorithm outside her institution has been difficult.
So she's collaborating with Apervita. “We spent years accumulating data, coming up with an algorithm, publishing that algorithm—and then to actually get it from that stage to (hospitals adopting it) turns out to be even more challenging,” she said. “It can take a very long time. And it's expensive.”
Apervita recently announced an $18 million funding round, including GE Ventures. The funding will be used to increase the number of developers and buyers on the market, Magelli said.
The healthcare algorithms Apervita shares are what he calls “computable,” meaning the protocol can be written into software code. While a top-notch clinician might be able to detect patient deterioration better than a software program, that person can't be everywhere and can't continuously monitor patients. Software analytics parsing data can monitor all the patients all the time, he said.
So one could create an algorithm, for example, that crunches patient data and predicts hospital readmissions. But institutions with such algorithms find it difficult to build and commercialize the software because of difficulties with distribution. “We're seeing an incredible rate of growth in computable content,” he said. But he's not seeing a conversion of that content into software and hopes his company can help solve that problem.
One technical challenge is getting a health system's clinical data and the software to “communicate.” The health system generally has to build an interface, which is expensive and laborious.
Magelli's solution was to build a cloud-based marketplace. Apervita offers standard templates for hospitals and other healthcare providers to upload their data and connect that data to analytics written by software developers based on a standardized, repeatable process. For each analytic, contributing providers may upload one or two streams of data related to a few vital signs or demographics. For example, on Apervita a customer can search for an algorithm on infection rates, scrutinize the evidence provided by the software developer and then apply that to its own population.